How big is the advantage, and where does that number come from?
The most directly usable estimate comes from classroom research rather than the lab. A review of applied studies — real curriculum material, real course timescales, more than 3,000 learners across 31 effect sizes — found a moderate advantage for distributed over massed practice, d = 0.54 [4]. In plain terms, a d of about 0.5 means the average distributed-practice learner lands around the middle of the massed-practice group's upper half: a genuine, practical difference, but not a transformation.
Second-language learning shows a similar picture. Across 48 experiments and 98 effect sizes, spacing produced a medium-to-large effect on language learning [1]. So two quite different bodies of evidence — classroom subjects and language learning — converge on "moderate to fairly large" rather than "huge." That consistency is reassuring, though both estimates come from meta-analyses whose abstracts report summary effects rather than letting us inspect individual studies.
Why does the advantage look bigger on later tests?
The central finding about timing is that spacing and retention interval work together rather than independently. In a synthesis of 839 assessments drawn from 317 experiments, the spacing interval that produced the best final retention grew as the retention interval grew [2]. In other words, there is no single "correct" gap between study sessions; the right gap depends on how long you need to remember the material.
Classroom evidence points the same direction, if more loosely. Studies with longer retention intervals generally showed larger effects of distributed over massed practice [4]. The authors caution that a full moderator analysis was not possible given the number of studies, so this is a pattern rather than a precisely quantified law. Still, it fits the laboratory picture: cramming can look fine on an immediate test and then fade, while spaced study holds up better later.
Does more spacing always beat less spacing?
Not necessarily, and the second-language evidence makes the distinction concrete. Shorter spacing was just as effective as longer spacing when students were tested immediately, but longer spacing was more effective on delayed tests [1]. That is the same joint effect seen in the laboratory synthesis, expressed as a practical rule: if the test is soon, a short gap is fine; if you need the knowledge to last, stretch the gap.
The same meta-analysis found that equal spacing and expanding spacing — where gaps get progressively longer — were statistically equivalent [1]. So the popular advice to always use an expanding schedule is not supported as a general rule here; what matters more is that study is spread out at all, and that the overall gap suits the retention interval.
When does spreading practice out fail to help?
The advantage is not universal, and the exceptions are instructive. A series of experiments on learning from texts found that distributed reading did not produce a delayed-test benefit, and in some cases hurt performance on immediate tests; the pattern held for single texts and for complementary multiple texts, and even at a two-week retention interval [5]. Notably, learners with low prior knowledge sometimes did worse when distribution was imposed [5].
This does not overturn the meta-analytic averages; it marks their boundary. The classroom and laboratory syntheses concern repeated study of the same material, where spacing reliably helps [2][4]. Reading different or complementary texts is a different task, and the dissertation's author frames those results as a first step rather than a verdict [5]. The practical reading: spacing is a strong default for memorization and skill practice, but it should not be assumed to transfer automatically to complex text comprehension, especially for students who lack background knowledge.
Sources used in this answer
The Effects of Spaced Practice on Second Language Learning: A Meta‐Analysis
Across 48 second-language experiments, spacing had a medium-to-large effect; shorter spacing matched longer spacing on immediate tests but was less effective on delayed tests, and equal and expanding spacing were statistically equivalent.
Distributed practice in verbal recall tasks: A review and quantitative synthesis.
In a synthesis of 839 assessments from 317 experiments, the spacing interval producing maximal retention increased as the retention interval increased.
A meta-analytic review of the benefit of spacing out retrieval practice episodes on retention
This meta-analysis examined the benefit of spacing out retrieval-practice episodes on retention, with test timing set by a specified retention interval.
The Distributed Practice Effect on Classroom Learning: A Meta-Analytic Review of Applied Research
In 31 classroom-study effect sizes covering more than 3,000 learners, distributed practice showed a moderate advantage (d = 0.54), with generally larger effects for longer retention intervals.
Improving Learning from Texts: Distributed Practice and Distributed Learning as Desirable Difficulty in Reading Single and Multiple Texts
In experiments on learning from single and multiple texts, distributed conditions showed no delayed-test advantage and sometimes worse immediate performance, with low-prior-knowledge learners sometimes impaired.